Association of white blood cell count with treatment response to cefepime versus piperacillin-tazobactam.

RATIONALE: Empiric anti-pseudomonal coverage is an important part of treatment of sepsis, but there are limited data to guide antibiotic selection. The Antibiotic Choice on Renal Outcomes (ACORN) randomized trial found no difference in mortality when comparing cefepime and piperacillin-tazobactam, while an instrumental variable (IV) study found improved mortality with cefepime.

OBJECTIVES

To determine if white blood cell (WBC) count modifies the effect of antibiotic choice on mortality.

METHODS

We conducted post hoc subgroup analyses of the ACORN trial and the IV study comparing cefepime to piperacillin-tazobactam. We performed regression modeling of interaction between treatment group and WBC on 28-day mortality with adjustment for relevant covariates, with evaluation for heterogeneity of treatment effect through likelihood ratio testing.

MEASUREMENTS AND MAIN

RESULTS

The main outcome was 28-day mortality in both cohorts. Secondary outcomes in the secondary analysis of the ACORN trial included highest stage of acute kidney injury on an ordinal scale, a composite outcome of major adverse kidney events, and incidence of coma or delirium within 14 days. There was a significant interaction between WBC and treatment group in predicting 28-day mortality in both the ACORN trial and the IV study (odds ratio [OR], 0.95 [95% CI, 0.92-0.98], P <.01 and OR, 0.95 [95% CI, 0.94-0.96], P <.01, respectively), with likelihood ratio testing revealing better model fit with inclusion of the interaction term (F = 8.17, P =<.01 and F = 42.9, P <.01, respectively). In the ACORN cohort, patients with a baseline WBC count of 16 or higher had significantly lower odds of mortality when treated with piperacillin-tazobactam compared to cefepime (OR, 0.51 [95% CI, 0.29-0.90]).

CONCLUSIONS

WBC modified the effect of antibiotic choice on mortality in 2 large clinical studies. Further research is needed to confirm this effect in prospective studies and to determine if WBC count could be used for predictive enrichment in future trials of anti-pseudomonal antibiotics.